| dc.description.abstract | Conventional multi-antenna transmitters use a separate RF chain (consisting of DACs, mixers, filters) and a power amplifier (PA) for each antenna element, and use modulation alphabets such as QAM/PSK for transmission. The consequences of this are, first, the RF hardware complexity, size, and cost increase with the number of antennas, and second, the linearity requirements for higher order QAM and OFDM transmissions affect the power efficiency of the amplifier in each RF chain. Load modulated array (LMA) is emerging as a promising multi-antenna transmission architecture that alleviates the aforementioned issues. An LMA uses a single central power amplifier (CPA) for the entire antenna array and no RF chains. LMA is based on the concept of load modulation, in which an antenna current proportional to the information bearing signal is achieved by modulating the antenna load impedances, while maintaining the amplifier input at a constant level. Varying the antenna load impedances with the information signal can cause a mismatch between the source impedance and the effective antenna impedance, causing power reflection into the CPA. A way to mitigate this is to ensure that the transmit signal in every channel use lies on the surface of a multidimensional hypersphere. This is called ‘phase modulation on the hypersphere (PMH)’. In this thesis, we investigate PMH signaling, detection, and precoding for LMAs in point-to-point, multiuser uplink, and multiuser downlink communication scenarios.
In the first part, we consider the construction of PMH signal vectors for LMAs. Construction of PMH signal vectors is typically non-analytic (e.g., clustering, potential maximization) and hence becomes computationally and storage wise expensive. We propose random phase modulation (RPM) as an inexpensive means of constructing PMH signal vectors. The idea of RPM is extended to random phase precoding (RPP) and precoder index modulation (PIM) to devise PMH signaling schemes that achieve good performance. Indexing in time and spatial domains is also investigated.
In the second part, we consider LMAs for communication in multiuser scenario. For multiuser communication on the uplink, we show using analysis and simulations that LMAs achieve superior bit error performance compared to other single RF chain multi-antenna transmission architectures in the literature. To exploit this performance advantage in large systems, we propose low complexity multiuser signal detection algorithms based on message passing and Monte Carlo sampling techniques. For multiuser communication on the downlink, we design a block diagonalizing precoder that nulls multiuser interference at the user terminals while ensuring that the precoded signal vector lies on the hypersphere. Further, when the antenna load impedances are tuned using discrete values from a finite set, the support of the precoded signal vector is finite. For this setting, we propose an iterative precoding algorithm using the generalized least square error (GLSE) framework.
In the final part, we propose a hybrid signaling scheme using LMAs and channel modulation (CM), wherein additional information bits are conveyed through the ON/OFF status of parasitics placed near the antenna array. For this scheme, we exploit the inherent sparsity in the signal vectors for multiuser detection on the uplink. We also incorporate CM in the GLSE framework for multiuser precoding on the downlink. | en_US |
